[llvm] [RISCV] Implement RISCVTTIImpl::shouldConsiderAddressTypePromotion for RISCV (PR #102560)

via llvm-commits llvm-commits at lists.llvm.org
Sun Aug 11 19:52:04 PDT 2024


LiqinWeng wrote:

> > > Why are there InstCombine changes in this patch?
> > 
> > 
> > I removed the optimization I had implemented in instcombine. I think that when the offset in the address calculation is a constant, it can be covered in the current patch and does not need to be implemented in InstCombine. ![image](https://private-user-images.githubusercontent.com/19768075/356495337-3a42f5d2-b714-4d4b-ac33-dfd7663b023c.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.Xjs9bSqme_QChppFuE0tkdTixrITWfL3kdTVZRDOf0A)
> 
> They should be separate PRs with different reviewers.

I restored the optimization of instcombine

https://github.com/llvm/llvm-project/pull/102560


More information about the llvm-commits mailing list